Freelance Software Engineer / LLM Trainer (ByteDance Seed Project)
Provided supervised fine-tuning (SFT) support by writing high-quality algorithmic and backend solutions used for LLM training. Designed coding problems and evaluation tasks to benchmark model reasoning and system-design capabilities. Built adversarial test cases and evaluation pipelines to improve robustness and edge-case handling.• Generated structured reasoning artifacts to improve model performance on coding and system-design tasks.• Authored and prepared training/evaluation code assets and workflows for reproducible experimentation.• Implemented Docker-based environments to ensure deterministic training and testing runs.